Files
netbird/proxy/internal/proxy/context.go
Maycon Santos b416063bcc [management,proxy] Agent network: per-account LLM gateway (policy, metering, multi-provider) (#6555)
* [agent-network] Shared proto, OpenAPI schema, and generated types

* [agent-network] Management: store, manager, synthesizer, policy engine, provider catalog, HTTP/gRPC API

Adds the account-scoped agent-network module: provider/policy/budget CRUD and
store, the reverse-proxy service synthesizer, policy selection + limit
enforcement, the provider catalog (incl. Vertex AI and AWS Bedrock entries),
and the management HTTP + proxy gRPC surfaces.

* [management] Fix agent-network proxy-peer fan-out on affected-peer recompute

The affected-peers resolver loaded only persisted reverse-proxy services, but
agent-network services are synthesized on demand and never persisted. As a
result the embedded proxy peer was never folded into the affected set when a
client's group changed, so the proxy received no network-map update for a newly
authorised client and rejected its handshake until a full resync (restart).

loadProxyServices now merges the synthesized agent-network services (injected
via a registration hook to avoid an import cycle), so proxy peers learn newly
authorised clients immediately.

* [proxy] Reverse-proxy middleware framework, chain, and request plumbing

The per-target middleware chain (slots, dispatcher, mutation gate, metadata
merger), body capture, access-log terminal sink, and the proxy wiring that
builds + runs chains for synthesized agent-network services.

* [proxy] LLM parsers, pricing, and builtin middlewares (OpenAI, Anthropic, Vertex AI, AWS Bedrock)

Request/response parsers and SSE/event-stream metering, the embedded pricing
table, and the builtin middleware set: request parser, router, policy
limit-check/record, cost meter, guardrail, identity inject, response parser.
Includes the path-routed providers — Google Vertex AI (keyfile:: service-account
OAuth minting) and AWS Bedrock (bearer auth, invoke/converse/streaming, optional
/bedrock prefix) — plus the Models allowlist and unmeterable-publisher deny.

* [proxy] IPv6 in-place apply and TCP accept-loop hardening on netstack listeners

* [agent-network] End-to-end test suite, module docs, and deployment preset

* [agent-network] Fix codespell typos and exclude false positives

- labelgen word pool: vermillion -> vermilion, racoon -> raccoon.
- codespell ignore list: add flate (Go compress/flate package), recordin
  (a test-local identifier), and unparseable (a valid alternative spelling used
  consistently across identifiers + a metadata-value constant).

* [management] Set LastSeen on injected proxy peer in realstack test (MySQL strict-mode)

The injected embedded proxy peer had a PeerStatus with a zero LastSeen, which
serializes to '0000-00-00' and is rejected by MySQL in strict mode (SQLite
tolerates it). Set LastSeen to a valid time so SaveAccount succeeds on both
engines.

* [agent-network] Remove e2e shell-script suite from this branch

The end-to-end shell scripts under scripts/e2e/ are maintained in a separate
testing suite and are not part of this change set.

* [agent-network] Polish module docs: remove internal review scaffolding, fix links, verify diagrams

Strip PR-review framing, commit references, absolute paths, and stale internal
references from the agent-network module docs; fix broken relative links; verify
all diagrams against the current architecture. Remove the internal AI-reviewer
prompt file.

* [management] Refine session expiration handling to support 3-state encoding for SSO deadlines

* [agent-network] Relocate agentnetwork package to internals/modules

Move management/server/agentnetwork (and its catalog/, labelgen/, types/
subpackages) to management/internals/modules/agentnetwork, alongside the
reverse-proxy module, and rewrite all importers. Pure relocation: package names,
the synthesizer + affectedpeers registration hook, and store access (shared
store.Store) are unchanged, so no import cycle is introduced (affectedpeers
still depends only on the agentnetwork/types leaf).

* [agent-network] Co-locate HTTP handlers in the module (RegisterEndpoints)

Move the agent-network HTTP handlers from server/http/handlers/agentnetwork into
the module at internals/modules/agentnetwork/handlers (package handlers) and
rename the entrypoint AddEndpoints -> RegisterEndpoints, matching the
reverse-proxy module convention. Wiring in http/handler.go updated accordingly.
2026-06-27 13:41:00 +02:00

305 lines
8.0 KiB
Go

package proxy
import (
"context"
"maps"
"net/netip"
"sync"
"github.com/netbirdio/netbird/proxy/internal/types"
)
type requestContextKey string
const (
capturedDataKey requestContextKey = "capturedData"
)
// ResponseOrigin indicates where a response was generated.
type ResponseOrigin int
const (
// OriginBackend means the response came from the backend service.
OriginBackend ResponseOrigin = iota
// OriginNoRoute means the proxy had no matching host or path.
OriginNoRoute
// OriginProxyError means the proxy failed to reach the backend.
OriginProxyError
// OriginAuth means the proxy intercepted the request for authentication.
OriginAuth
)
func (o ResponseOrigin) String() string {
switch o {
case OriginNoRoute:
return "no_route"
case OriginProxyError:
return "proxy_error"
case OriginAuth:
return "auth"
default:
return "backend"
}
}
// CapturedData is a mutable struct that allows downstream handlers
// to pass data back up the middleware chain.
type CapturedData struct {
mu sync.RWMutex
requestID string
serviceID types.ServiceID
accountID types.AccountID
origin ResponseOrigin
clientIP netip.Addr
userID string
userEmail string
userGroups []string
// userGroupNames pairs positionally with userGroups; populated from
// the JWT's group_names claim or from ValidateSession/Tunnel
// responses. Slice may be shorter than userGroups for tokens minted
// before names were resolvable.
userGroupNames []string
authMethod string
metadata map[string]string
agentNetwork bool
suppressAccessLog bool
}
// NewCapturedData creates a CapturedData with the given request ID.
func NewCapturedData(requestID string) *CapturedData {
return &CapturedData{requestID: requestID}
}
// GetRequestID returns the request ID.
func (c *CapturedData) GetRequestID() string {
c.mu.RLock()
defer c.mu.RUnlock()
return c.requestID
}
// SetServiceID sets the service ID.
func (c *CapturedData) SetServiceID(serviceID types.ServiceID) {
c.mu.Lock()
defer c.mu.Unlock()
c.serviceID = serviceID
}
// GetServiceID returns the service ID.
func (c *CapturedData) GetServiceID() types.ServiceID {
c.mu.RLock()
defer c.mu.RUnlock()
return c.serviceID
}
// SetAccountID sets the account ID.
func (c *CapturedData) SetAccountID(accountID types.AccountID) {
c.mu.Lock()
defer c.mu.Unlock()
c.accountID = accountID
}
// GetAccountID returns the account ID.
func (c *CapturedData) GetAccountID() types.AccountID {
c.mu.RLock()
defer c.mu.RUnlock()
return c.accountID
}
// SetOrigin sets the response origin.
func (c *CapturedData) SetOrigin(origin ResponseOrigin) {
c.mu.Lock()
defer c.mu.Unlock()
c.origin = origin
}
// GetOrigin returns the response origin.
func (c *CapturedData) GetOrigin() ResponseOrigin {
c.mu.RLock()
defer c.mu.RUnlock()
return c.origin
}
// SetClientIP sets the resolved client IP.
func (c *CapturedData) SetClientIP(ip netip.Addr) {
c.mu.Lock()
defer c.mu.Unlock()
c.clientIP = ip
}
// GetClientIP returns the resolved client IP.
func (c *CapturedData) GetClientIP() netip.Addr {
c.mu.RLock()
defer c.mu.RUnlock()
return c.clientIP
}
// SetUserID sets the authenticated user ID.
func (c *CapturedData) SetUserID(userID string) {
c.mu.Lock()
defer c.mu.Unlock()
c.userID = userID
}
// GetUserID returns the authenticated user ID.
func (c *CapturedData) GetUserID() string {
c.mu.RLock()
defer c.mu.RUnlock()
return c.userID
}
// SetUserEmail records the authenticated user's email address. Used by
// policy-aware middlewares to stamp identity onto upstream requests
// (e.g. x-litellm-end-user-id) without a management round-trip.
func (c *CapturedData) SetUserEmail(email string) {
c.mu.Lock()
defer c.mu.Unlock()
c.userEmail = email
}
// GetUserEmail returns the authenticated user's email address. Returns
// the empty string when the auth path didn't carry an email (e.g.
// non-OIDC schemes or legacy JWTs minted before the email claim).
func (c *CapturedData) GetUserEmail() string {
c.mu.RLock()
defer c.mu.RUnlock()
return c.userEmail
}
// SetUserGroups records the authenticated user's group memberships so
// downstream policy-aware middlewares can authorise the request without
// an additional management round-trip. The auth middleware populates this
// from ValidateSessionResponse / ValidateTunnelPeerResponse and from the
// session JWT's groups claim on cookie-bearing requests.
func (c *CapturedData) SetUserGroups(groups []string) {
c.mu.Lock()
defer c.mu.Unlock()
if len(groups) == 0 {
c.userGroups = nil
return
}
c.userGroups = append(c.userGroups[:0], groups...)
}
// SetAgentNetwork records whether the request hit a synthesised
// agent-network target. The terminal access-log middleware stamps the
// flag onto the proto so management can distinguish synthetic traffic.
func (c *CapturedData) SetAgentNetwork(b bool) {
c.mu.Lock()
defer c.mu.Unlock()
c.agentNetwork = b
}
// GetAgentNetwork reports whether the request matched a synthesised
// agent-network target.
func (c *CapturedData) GetAgentNetwork() bool {
c.mu.RLock()
defer c.mu.RUnlock()
return c.agentNetwork
}
// SetSuppressAccessLog records whether the per-request access-log emission
// must be skipped for this request. Stamped from the matched target's
// DisableAccessLog flag so the access-log middleware can short-circuit
// log delivery for opted-out agent-network targets.
func (c *CapturedData) SetSuppressAccessLog(b bool) {
c.mu.Lock()
defer c.mu.Unlock()
c.suppressAccessLog = b
}
// GetSuppressAccessLog reports whether access-log emission has been
// suppressed for this request.
func (c *CapturedData) GetSuppressAccessLog() bool {
c.mu.RLock()
defer c.mu.RUnlock()
return c.suppressAccessLog
}
// GetUserGroups returns a copy of the authenticated user's group
// memberships.
func (c *CapturedData) GetUserGroups() []string {
c.mu.RLock()
defer c.mu.RUnlock()
if len(c.userGroups) == 0 {
return nil
}
out := make([]string, len(c.userGroups))
copy(out, c.userGroups)
return out
}
// SetUserGroupNames records the human-readable display names for the
// user's groups, ordered identically to UserGroups (positional
// pairing). Stamped onto upstream requests as X-NetBird-Groups so
// downstream services can read names rather than opaque ids.
func (c *CapturedData) SetUserGroupNames(names []string) {
c.mu.Lock()
defer c.mu.Unlock()
if len(names) == 0 {
c.userGroupNames = nil
return
}
c.userGroupNames = append(c.userGroupNames[:0], names...)
}
// GetUserGroupNames returns a copy of the authenticated user's group
// display names. Position i pairs with UserGroups[i]. May be shorter
// than UserGroups for tokens minted before names were resolvable; the
// consumer should fall back to ids for missing positions.
func (c *CapturedData) GetUserGroupNames() []string {
c.mu.RLock()
defer c.mu.RUnlock()
if len(c.userGroupNames) == 0 {
return nil
}
out := make([]string, len(c.userGroupNames))
copy(out, c.userGroupNames)
return out
}
// SetAuthMethod sets the authentication method used.
func (c *CapturedData) SetAuthMethod(method string) {
c.mu.Lock()
defer c.mu.Unlock()
c.authMethod = method
}
// GetAuthMethod returns the authentication method used.
func (c *CapturedData) GetAuthMethod() string {
c.mu.RLock()
defer c.mu.RUnlock()
return c.authMethod
}
// SetMetadata sets a key-value pair in the metadata map.
func (c *CapturedData) SetMetadata(key, value string) {
c.mu.Lock()
defer c.mu.Unlock()
if c.metadata == nil {
c.metadata = make(map[string]string)
}
c.metadata[key] = value
}
// GetMetadata returns a copy of the metadata map.
func (c *CapturedData) GetMetadata() map[string]string {
c.mu.RLock()
defer c.mu.RUnlock()
return maps.Clone(c.metadata)
}
// WithCapturedData adds a CapturedData struct to the context.
func WithCapturedData(ctx context.Context, data *CapturedData) context.Context {
return context.WithValue(ctx, capturedDataKey, data)
}
// CapturedDataFromContext retrieves the CapturedData from context.
func CapturedDataFromContext(ctx context.Context) *CapturedData {
v := ctx.Value(capturedDataKey)
data, ok := v.(*CapturedData)
if !ok {
return nil
}
return data
}